Visual Analytics Summer School (CANVAS) 2014 Program
To register for Web access to view the keynotes and presentations using GoToWebinar, click on the title for that presentation.
Monday, July 28, 2014
Fred Popowich (VIVA/SFU) - Status of VA in Canada.
Sara Diamond (OCADU) -Representing data: Visualization, sonification, data sculpting and vibro-tactile arrays.
Abstract: This presentation will discuss some of the fundamentals of design knowledge and methods for data visualization and explain the relevance of design to visual analytics. Included are basic design approaches to data and methods of working with clients and user groups from a design perspective. Having established some basics, I will explore some of the emerging fields of data representation and discuss the relevance of these to data types and contexts. These include sonification, haptic interfaces, vibro-tactile displays and data sculpture. Examples will be provided.
CANVAS Keynote: John Stasko (GeorgiaTech) - Visual analytics for investigative analysis and exploration of documents and data.
Abstract: Whether investigators are fighting crime, curing diseases, deciding what car to buy, or researching a new field, inevitably they will encounter text documents. Unfortunately, plain (unstructured) text documents are difficult to analyze and understand, especially large collections of documents. The field of visual analytics holds promise for helping investigators with such problems. Visual analytics combines computational data analysis with interactive visualization in the context of understanding how people think and reason. It can be particularly effective in situations when the data is large and unfamiliar, and the analyst must browse and explore to learn about a situation or domain. In this talk I will describe principles from the field, illustrating how visualizations help people make sense of data.
Nathalie Riche (Microsoft) - Visual exploration of complex data.
Abstract: As we enter the era of big data, researchers from many fields strive to help data scientists capture, process, store, search, and analyze their data. While the sheer amount of data raises new research challenges, data scientists and enthusiasts alike already wrestle with the complex nature of “small” datasets that they collect. In this talk, I will focus on visual techniques to explore complex data. I will demonstrate how interactive visualization can play a crucial role in helping people raise new questions about their data and rapidly form hypotheses.
1330-1630 Schedule A
VARI-Lab Exercises for 24 CANVAS Students
1330-1630 Schedule B
CANVAC Challenge Panel I – Re-Thinking Visual Analytics in a World of Big Data, Information Visualization, and Data Analytics.
Abstract: In 2005, Thomas & Cook defined Visual Analytics (VA) as the science of analytical reasoning facilitated by interactive visual interfaces. VA tools and techniques enable analysts to explore and synthesize information and derive insight from massive, dynamic, ambiguous, and possibly conflicting data; detect the expected and discover the unexpected; provide timely, defensible, and understandable assessments; and communicate assessment effectively for action.
At that time, Thomas & Cook proposed a number of challenges:
- Address the challenges and seize the opportunities posed by the large scale of the analytic problems and data available.
- Create a science of visual representation and a suite of visual paradigms based on cognitive and perceptual principles that can be deployed through engineered, reusable components to support the analytical reasoning process.
- Develop a science of interactions and taxonomy of interaction techniques that support the analytical reasoning process.
- Develop data theory and practice for transforming data, synthesizing different data representations and representing data quality, reliability and certainty that engenders trust.
- Create visually enabled tools to support collaborative analytic reasoning.
- Develop methods and tools to capture the analytic assessment process and to communicate analytic results, decisions and recommendations
- Are this VA definition and its challenges still appropriate in the current world of visual analytics, interactive analytics systems, information visualization, data analytics and Big Data?
- How should VA fit into this new world?
Ronald Rensink (UBC, Moderator)
Brian Fisher (SFU)
John Stasco (Georgia Tech)
Nathalie Riche (Microsoft Research)
Anoop Sarkar (SFU)
Sheelagh Carpendale (UCalgary)
CANVAS Reception – Forestry Building, UBC
Tuesday, July 29, 2014
Evangelos Milios (DalhousieU) - Visual text analytics: Facilitating insight from text data. (remote from Dalhousie U)
Abstract: Text data is big and growing. It includes social media, scientific research papers, forums, corporate documents, and government text. Knowledge is being captured and made available through online resources such as Wikipedia. Often text corpora are networks of documents, such as the citation network. How can users of text resources make sense of them, if they are not text mining experts? Visual text analytics aims to connect text mining and text visualizations to enable the sense-making task. In this tutorial, we will review basic visualization techniques and supporting text mining methods, including text clouds, topic models, text streams, sentiment visualization and multi-view systems, which integrate multiple interconnected visualizations of the same text data from different points of view. We will present case studies that illustrate the interaction of the user with the text mining and visualization components of multi-view systems. We will include pointers to the relevant literature to enable further study.
Ronald Rensink (UBC) - Visual analytics and visual intelligence.
Abstract: Over the past few decades, several major breakthroughs have occurred in our understanding of human vision, and in particular, in our understanding of visual intelligence. This talk reviews some of what has been done, and discusses several recent developments, including the discovery of general laws on the perception of correlation.
Academic Keynote: Leland Wilkinson (SkyTree) - Exploring huge collections of scatterplots and images.
Abstract: We introduce a method and application for guiding interactive exploration of a huge corpus of scatterplots. The method is based on scagnostics - nine characterizations of the 2D distributions of orthogonal pair-wise projections on a set of points in multidimensional Euclidean space. These characterizations include measures such as, density, skewness, shape, outliers, and texture. Working directly with these measures, we can locate anomalies for further analysis or search for similar distributions in a giant scatterplot matrix. We also present an extension of this technique that can be applied to huge collections of images (joint work with Tuan Dang).
Sheelagh Carpendale (U Calgary) - The interplay between representation and interaction.
Abstract: Much of the excitement in the early 1990s about information visualization originated in the idea of creating new visual, spatial representations that would allow people to ‘see’ their data. Much was said about the amount of the brain that is devoted to spatial and visual reasoning and how visualizations might have the power to utilize these relatively untapped resources. However, as information visualization research has progressed a degree of practically has emerged – heightening a focus on usability and task enablement. As important as this focus may be, there may still be something worth investigating in the notion of alternate representations. In this talk, I will explore the possible power of alternate interactive visual representations by considering both ideas around innovation and practical illustrations.
1330-1630 Schedule A
VARI-Lab Exercises for 24 CANVAS Students
1330-1630 Schedule B
CANVAC Challenge Panel II – Visual Analytics Research: Moving Theory Into Practice & Research Into Industry.
Abstract: Moving research into practice is one of the goals necessary to successfully leverage multi-disciplinary visual analytics research (Thomas & Cook, 2005). To do this, they suggested the following actions to support VA research:
- Develop VA research methods and tools to facilitate evaluation of new VA concepts and technologies; develop common component-based software tool-sets.
- Create a common security and privacy infrastructure to provide access to appropriate data.
- Identify and publicize best practices for deploying and using VA technologies in operational environments.
- Have we succeeded in moving VA research into practice and industry? Do we have compelling case studies that can be used to promote visual analytics research and application?
- How can we successfully conduct both basic and applied VA research? What infrastructure and processes are available, or necessary, to enhance visual analytics research and application?
David Darvill (SFU, Moderator)
Leland Wilkinson (SkyTree)
Orland Hoeber (URegina)
Christopher Collins (UOIT)
David Kasik (Boeing)
Wednesday, July 30, 2014
Orland Hoeber (URegina) - Supporting search with visual analytics.
Abstract: Searching within unstructured or semi-structured document collections has become a common activity in our modern information-centric society. These collections range in size and complexity, examples of which include the web, online digital libraries, corporate document collections, image collections, social media, email, and the files on a personal computer. While simple fact verification is well supported by current search technologies in some contexts (i.e., searching the web), when the information seeking tasks and goals become more complex, a substantial cognitive burden is placed on the searchers to craft and refine their queries, evaluate and explore among the search results, and ultimately making sense of what is found. This burden often stems from searchers' incomplete knowledge regarding their information seeking goals, their inability to formulate accurate queries, and their low tolerance for considering the relevance of the search results. Visual analytics provides a means for supporting and enhancing the search process, through a combination of automatic machine learning approaches and interactive visualization methods. The goal is to automatically extract and infer relevant information during the search activity and present this to the searchers in a visual format that allows for quick interpretation and easy manipulation of the information, providing support for a broad range of search goals and intentions.
In this Canadian Visual Analytics Summer School session, a number of important factors in the design, development, and study of visual analytics methods to support search tasks and goals will be explored, including issues associated with automatic information extraction, visual encoding, task-centric support, and evaluation methods. These will be discussed in the context of a series of visual search analytics systems, followed by a discussion on directions for future research.
Christopher Collins (UOIT) - Semantics and Sentiment in Visual Text Analytics.
Abstract: How do people feel about my product? What are the main themes in the news today? These are examples of the questions people ask about large scale text data. Visual text analytics tools are being created to help address these challenging questions. In this talk I will review recent research advances for exploring and analyzing sentiment in text, and extracting meaning and relationships between entities in text. For those visualization designers who want to take advantage of semantics and sentiment, this session will also cover natural language processing toolkits and data resources.
Academic Keynote: Wolfgang Stuerzlinger (York) - Interactive alternatives in visual analytics.
Abstract: I present a new project to enable interactive alternatives in Visual Analytics. The new solutions support the iterative and reflective nature of Visual Analytics by enabling users to speculatively generate and evaluate many alternatives. Users can re-visit previously explored solutions and even import processes they have developed within other alternatives. At a technical level, the system will enable individual or parallel editing of alternative analyses, support re-synchronization of alternatives after individual changes, and provide for the quick generation of new alternatives.
Margaret-Anne Storey (UVic) - Visualization for software analytics.
Abstract: The popularity of software visualization research over the past 30 years has led to innovative techniques that are now seeing widespread adoption by professional software practitioners. But this research has barely kept pace with some of the radical changes occurring in software engineering. In this talk, I first provide a retrospective on the field of software visualization and discuss some of the prominent techniques developed by research and adopted by industry, highlighting some lessons learned. Next I explore current trends in software engineering, including the prevalence of software ecosystems and software delivery as a service, and the emergence of the social coder within a participatory development culture. I will also discuss how the field of software analytics has matured and seeks to support practitioners in improving software quality, user experience and developer productivity through data-driven tasks. Finally, I suggest that software visualization should be playing a bigger role in these recent trends, emphasizing that interactive visualizations are poised to play a critical role in the field of software analytics.
1330-1630 Schedule A
VARI-Lab Exercises for 24 CANVAS Students
1330-1630 Schedule B
CANVAC Challenge Panel III – Visual Analytics Education, Training and Certification.
Abstract: There will be a growing need for knowledgeable personnel, i.e., Data Scientists, who can deal effectively with all of the elements of complex analytics environments that will include visual analytics, information visualization, data analytics, statistics, data mining, Big Data, etc.
- Academic programs at the undergraduate and graduate levels will be required to educate visual analytics professionals. How can this need be satisfied within Canada’s post-secondary institutes? How can multi-disciplinary programs for visual analytics be developed? What programs are currently available?
- Training of visual analytics professionals who are currently working in Canadian industry will be needed to ensure that visual analytics is effectively deployed into organizations. How should such training or re-training be developed and delivered? What courses are currently available?
- As an applied profession, is there a need for certification of visual analytics professionals? If so, who should offer such certification and how should it be done?
Fred Popowich (SFU, Moderator)
Sara Diamond (OCADU)
Christopher Rogers or Theo Rosenfeld (AeroInfo Systems)
Derek Reilly (DalhousieU)
Jean- Sébastien Mercier (VIVA)
CANVAS Pub Night – Point Grill, UBC
Thursday, July 31, 2014
Visual Analytics Industry Workshop I
Panel 1: VA Tool Vendors and VA Tool Users
Abstract: The objective of this panel is to identify the gaps and areas of intersection or convergence between what tool vendors are developing in the area of VA and what analysts and practitioners are using and need from those tools.
The purpose of this panel is to contribute to the alignment of user objectives with product development requirements.
Jean-Sebastien Mercier (VIVA, Moderator)
Michael O’Connell (TIBCO)
Rob Wilson (SAS)
Ian Prinsloo (ICBC)
Jacob Kuijpers (Deloitte)
Michael O’Connell (TIBCO) - Overview of TIBCO Spotfire.
Government I: Valérie Lavigne (Defence R&D Canada - Valcartier) - Visual analytics research at DRDC: Maritime domain awareness, social network analysis and semantic text exploration.
Abstract: Defence Research and Development Canada (DRDC) supports defence and security operations at home and abroad with knowledge and technology. Over the last years, DRDC has performed applied research to explore the potential of visual analytics science and technology in a number of domains. First, we explored visual representations than can make maritime anomalies salient and enable better analysis of a vessel-of-interest. Second, we applied visual analytics concepts to explore the results of social network analysis in a counterinsurgency context. Finally, we used lexical semantic models to improve the exploration of content from text to support more efficient intelligence analysis.
BRAVA Visual Analytics Workshop
BRAVA Overview – tbd
Júnia Anacleto (UFSCar) - tbd (Remotely)
Rodger Lea (UBC) - tbd
Sara Diamond, Hudson Pridham & Bhuvaneswari Arunchalan (OCAD University) - Monitoring Care and Condition
Abstract: We discuss the work undertaken in the context of the CAIS hospital in Brazil. We present a work-in-progress prototype for a tablet-based visual analytics tool that captures, structures and visualizes informal healthcare data as well as caregiver networks and helps visualize change in a resident’s condition over time through a caregiver voting process. The visualization is designed to capture and structure two principal categories of health information: condition and care. There is a circular relationship between care and condition: a resident’s condition informs the care and treatment they need, which, in turn, should impact their condition, and so on over time.
Long term health care requires integrated 24/7 care, communication and collaboration among a diverse team of general health practitioners, specialists, therapists, pharmacists, nurses, support providers, and even friends and family. For many of these practitioners, care is demanding, physical, hands-on and mobile work that makes real time record keeping difficult. In recent years, the medical profession has been transitioning to electronic medical records (EMR), but with mixed results. Marshall McLuhan gave us the idea that we shape our tools and thereafter our tools shape us. In other words, the technologies and media we use impact not only their content but also how we organize ourselves thereafter. Our tablet-based prototypes have tried to leverage informal data and care networks in a way that traditional medical records typically do not. But more than that, they also have implications for the organization of care. By giving every caregiver a ‘vote’ it potentially levels the playing field in what is traditionally a very hierarchical profession.
Derek Reilly (Dalhousie U) - Natural interfaces for accessing and sharing visually rich, information dense medical documents.
Abstract: Graphic design and information visualization are core to supporting analytics in complex mobile, collaborative workflows. In health care, patient information is accessed and shared within a complex framework of national and institutional rules, which are manifested in group and individual behaviours. Both at-a-glance, just-in-time visual interfaces and exploratory interactive visualizations must be considered within this framework. In this talk I will discuss our exploration of techniques for augmenting physical privacy-related actions (orienting documents, selecting and hiding documents, etc.) with digital support for privacy-sensitive document sharing.
BRAVA Project Panel
CANVAS 2014 Dinner – Point Grill, UBC
Friday, August 1, 2014
Visual Analytics Industry Workshop II
Panel 2: VA Tool Vendors and Universities
Abstract: The objective of this panel is to identify the areas of intersection / convergence between what VA tool vendors are developing and the research in the science of visual analytics being conducted in universities.
The purpose of this panel is to foster and facilitate joint research projects among tool vendors and academics researchers.
Robert Grace (Mitacs, Moderator)
Stephen Perelgut (IBM)
Rock Leung (SAP)
Mark Salopek, GRAND NCE
Alex Razoumov (Compute Canada, WESTGRID)
Rob Wilson (SAS) - Big Data, Big Insight
Abstract: In this seminar you see how you can apply today’s latest software and data visualization capabilities to address your specific business needs and interests. With a laser focus on SAS Visual Analytics, you’ll discover how easy it is to manage complex, massive data sets and quickly create and share dynamic visuals. You’ll also learn about on-the-fly forecasting, autocharting, "what does it mean?" pop-ups, and drag-and-drop capabilities.
Industry I: Alan Keahey (IBM) - Visualization scalability for Big Data.
Abstract: The human visual perception system is the highest bandwidth channel that we have for getting information into our brains. It provides unique features of instant feature and pattern recognition that scale well to large human-scale datasets. However, the scalability of our visual display and perception is not infinite, and the limits of it are reached well before we reach the sizes seen in typical Big Data applications. This talk will explore some of these differences in scale, and show how a variety of visual and analytical tools can be used to help bridge the scale gap between what we can see and the data that we need to understand.
Industry II: Ahmad Yaghoobi (Boeing Analytics) - Analytics at Boeing: Application of analytics, e.g., visual, text, predictive, to industry problems and challenges.
Abstract: Applications of advanced analytics to industrial problems in aerospace range from maintenance & reliability, material sciences, workplace safety, airplane safety, supply chain management to factory floor quality. These challenges require addressing such issues as defining the problem, identifying, acquiring, cleaning, and structuring data from multiple sources and formats, and, most importantly, communicating the results. In this presentation, we will discuss the identification & application of appropriate analysis methods, e.g., paired-analysis, and tools, e.g., Tableau, IN-SPIRE, Starlight, SPSS, SAS, R and Python, and the challenges of finding resources with the required skills, technology, and ability to perform analysis while managing stakeholder expectations.
Panel 3: VA Tool Users and Universities
Abstract: The objective of this panel is to identify gaps between Visual Analytics training and education programs and the needs of industry for High Quality Personnel who possess VA skills.
The purpose of this panel is improving academic programs in VA, data analytics and Big Data by better understanding the needs of Industry.
Iain Begg (VIVA, Moderator)
Gordon Hamilton (TDWI)
Kevin Ho (Engage Data)
Ron McMillan (BCIT)
Tamara Munzner, (UBC)
CANVAS Postscript: David Kasik (Boeing) - What we’ve learned.